Methodology for detection of ERD/ERS EEG patterns produced by cut events in film fragments

Publikation: Bidrag til bog/antologi/rapportKonferencebidrag i proceedingsForskning

The goal of this communication is to create a framework to isolate the neural reactions registered through EEG as a consequence of a specific input, among all those caused by an audiovisual. In order to do that, we analysed the neuronal register of the power change reactions related to specific cinematographic techniques, in this case the shot change by cut.

To research the shot change by cut through the neuronal record, one could use ad hoc audiovisual material specifically made for the experiment were the inputs are artificially introduced, or one could work with commercial cinematographic material, in order to have a more ecological approach For the latter approach, a more complex signal analysis process is needed because in the neuronal register the cut reactions are difficult to isolate.

For this experiment we used the EEG records of 21 subjects watching fragments extracted from 4 feature films that represent different styles and technical approaches. From the 21 user’s records we created a model signal for each film and compared the power change between the different model signals of each films through permutations test, Spearman correlation and analysis of slopes. Through an automated process with sliding time windows, we were able to locate those temporal lapses, electrodes and frequency bands that show reactions to the shot change by cut in the power change that have synchronous reactions in all the model signals. We also located the neuronal reactions that suppose representative variations in the ERD/ERS due to the cut input.
OriginalsprogEngelsk
TitelAugmented Cognition : Theoretical and Technological Approaches. HCII 2020
RedaktørerDylan D Schmorrow, Cali M Fidopiastis
Antal sider19
UdgivelsesstedCham
ForlagSpringer
Publikationsdato2020
Sider157-175
ISBN (Trykt)978-3-030-50352-9
ISBN (Elektronisk)978-3-030-50353-6
DOI
StatusUdgivet - 2020
Eksternt udgivetJa
BegivenhedInternational Conference on Human-Computer Interaction - Copenhagen, Danmark
Varighed: 19 jul. 202024 jul. 2020
Konferencens nummer: 22

Konference

KonferenceInternational Conference on Human-Computer Interaction
Nummer22
LandDanmark
ByCopenhagen
Periode19/07/202024/07/2020
NavnLecture Notes in Computer Science
Vol/bind12196
ISSN0302-9743

Bibliografisk note

(Ekstern)

Links

ID: 315538796